K Means Clustering

• Clustering is the task of taking data and generating clusters from it.
• It's an unsupervised task.
• K Means (pg. 753) is a simple clustering algorithm.
• Given a set of data points, pick K cluster centers randomly.
• Go through all of the data points and put them in the cluster that they are nearest to. (How do you calculate nearness?)
• Move the K cluster points to the center of their cluster. (How do you calculate the center of the cluster?)
• Repeat until no data points move clusters.
• Here you need to pick the number K, but otherwise the algorithm does all the work.